Modeling Roth Conversions

My wife and I would like to convert $10,000 each per year from our regular IRAs to our Roth IRAs from now (we are in our early 60's) until we each reach 70. Since we are currently retired, we will be making no other contributions to our retirement accounts aside from the Roth conversions.

To model that in ESP, would we: 1) each enter $10K per year in the Roth column until age 70 on the Retirement Accounts/Contributions tab? and 2) each enter in the Retirement Accounts/Special Withdrawals tab $10K above the baseline annual retirement account withdrawal recommended in the Details/Retirement Accounts section of the report (e.g if recommended withdrawals are $60K without Roth conversions, input $70K per year special withdrawals from now until age 70)?

Re: Step 2, I first tried entering just $10K per year in the Special Withdrawals tab, thinking that only $10K each counted as a "special" withdrawal, but then ESP seemed to assume that our total aggregate retirement account withdrawals for those years was only $10K per person, rather than the $10K conversion, plus whatever we needed to withdraw to live on.


dan royer's picture

Yes, you got it right. Indeed, that special withdrawal entry is really a kind of override. So you are correct: generally add the "special" amount to the already observable smooth amount. The withdrawals of course will incur tax as you expect, but the Roth contributions then just add to the Roths. The setting then too is to take Roth last.


dan royer's picture

I'm always curious how much this helps. Some cases seems to be more impacted than others.

Thanks for the swift response, Dan.

We're in a low tax bracket now, which will increase when SS kicks in (probably at age 70), so paying the tax now at a lower rate will help us by several thousands of dollars per year in Living Standard Per Adult.

I have noticed, though, that doing the gradual Roth conversions caused the Monte Carlo results to be much more noisy, and in several scenarios, they have predicted 0 income from our early 70's on, but in the median and high returns columns, not in the low or very low returns columns as one might expect. This disappears when I take out the gradual Roth conversions.